It seems redundent to have both dbmail and the MTA check for the validity of the domain name and users. I can understand the problems with spam relays, but can't you just make an MTA that will not deliver mail outside of dbmail? An MTA that only delivers to dbmail would be smaller and quicker. Or would this be against the SMTP protocal. From a solutions provider perspective who really only cares about the people with paid accounts on their system, it seems logical to not even bounce mail. Partially because I think spamer's can use the method to differentiate between real addresses, but also because it takes up resources that don't benefit the customer.
It seems to me that the only required and useful purpose of an MTA is to pipe the data to dbmail. What other purposes does a MTA have? Could someone just write a simple perl script that binds to the smtp port and accepted connections and pipes the data to dbmail-smtp? >From a newbie's perspective to dbmail, the most challenging part to installing >dbmail is figuring out how to interface it with your current MTA. I'm >convinced that support would be easier if a simple dbmail MTA was provided in >the distribution. Status: O Matt btw, [EMAIL PROTECTED] is part of a spam filter email address, one which I am trying to emulate with dbmail. >From Jeff Brenton <[EMAIL PROTECTED]> on 5 Oct 2003: Status: O > Hello dbmail, > > dmc> My question is, is there any MTAs that will simply pass the mail > dmc> onto dbmail-smtp without checking to see if the domain is valid? > dmc> Or is there a way I can define sendmail to accept all mail that > dmc> it is sent? Would there be a downside to doing this? My > dmc> understanding is that DBMail would still bounce the email if it > dmc> did not match a domain in the database. > > You don't want to have the MTA accepting any mail that it doesn't have > to. With Postfix, you can simply put the transport table into a mysql > database, and have Postfix use it to determine whether or not it is > the destination for a domain, and how to handle it once it accepts it. > > I use a table called "mytransport", which contains records like: > > espi.com dbmail: > mydomain.com dbmail: > > Postfix has these parameters: > > mydestination = mysql:/etc/postfix/my_trans.cf > transport_maps = mysql:/etc/postfix/my_trans.cf > > my_trans.cf says: > > user = postfix > password = password > dbname = forwarder > table = mytransport > hosts = 192.168.0.2 > select_field = transport > where_field = domain > > For each inbound, Postfix does the lookup to see if the domain is > returned or no rows... If it is acceptable, it then does the same > lookup to determine how to handle the domain's mail, which is dbmail > in these cases. > > Similarly, it uses the dbmail aliases table to determine if the > destination address actually exists, so bounces are generated at the > MTA, rather than with dbmail. > > Why is this last item important? Because any MTA that accepts mail it > can't deliver is an invitation to abuse by spammers. They send mail > destined for "[EMAIL PROTECTED]" to your mail server, showing a from > address of "[EMAIL PROTECTED]". Your machine accepts, figures out > that it can't deliver it, and bounces it... to "[EMAIL PROTECTED]", > rather than the spammer. > > -- > Best regards, > Jeff mailto:[EMAIL PROTECTED] > > _______________________________________________ > Dbmail mailing list > [email protected] > https://mailman.fastxs.nl/mailman/listinfo/dbmail >
